The work aims to design an improved salp swarm algorithm to solve the model constructed, and verify the validity and practicability of the model and the algorithm. A multi-objective model aiming to minimize total transport costs and carbon emissions while reducing risks was established. The model was solved using a salp swarm algorithm that incorporated chaotic mapping, information sharing mechanism and multiple swarm strategies. Through random examples, actual case verification and comparison with the basic salp swarm algorithm, it could be seen that the improved salp swarm algorithm had superior optimization performance. Solving the low-carbon cold chain multimodal vehicle routing model with the improved salp swarm algorithm can improve efficient solutions, which provide effective strategies for decision makers to face the multi-objective decision problem, and help to formulate better vehicle routing in practical applications.